Smart-Money Mirror v1 — executable strategy

First-track paper deployment that mirrors top-N positions across our 7 tracked managers (Druckenmiller, Marathon, Berkshire, Russo, Akre, Tiger Global, Tepper). Deploys NOW because the /investing:smart-money-watch skill is already built and the 2026 Q1 13F backfill is on disk. The lag (45-90 days from quarter-end disclosure) is the cost; the access to their selection IS the alpha. This strategy is intentionally simple — that's the point.

Honest framing

This is NOT a strategy that out-thinks Druckenmiller, Berkshire, or Tiger Global. It IS a strategy that copies positive-sum managers who already out-think markets, with a 45-90 day disclosure lag. The 13F filing window is what we pay for the read. The selection itself — what these allocators chose to size up on, after they did the work we cannot replicate — is the alpha.

Tracked manager set (7, Pabrai excluded per 13F lookup failure in 2026-05-17 backfill):

Selection rules (deterministic, not narrative)

1. Persistent darlings (high-confidence core)

Definition: any name in top-10 across 3+ managers AND held for 2+ consecutive quarters.

Per the [[01-projects/investing/anchors/smart-money/2026-05-17-aggregate-2yr-backfill-summary]] persistent-darlings table:

Allocation: 0.5R each = $2,500 × 4 = $10,000.

(Original draft was 1R each = $20k on persistent alone, which consumed the full initial budget. Re-spec to 0.5R each leaves room for signal-track adds + cash reserve.)

2. Recent multi-manager additions (signal-track)

Definition: any name where 2+ managers added or significantly increased same quarter, AND the add was within the last 2 quarters (currently 2025 Q4 + 2026 Q1).

Per the aggregate brief's collective-additions table for 2026 Q1:

Allocation: 0.5R each = $2,500 × 2 = $5,000.

(SNDK + INTC also overlap with the memory-cycle v1.1 thesis basket. That overlap is deliberate — when two independent signals — a structural thesis AND multi-manager smart-money corroboration — point at the same name, both signals get their own paper trade. Position-tracking ledgers stay distinct per strategy.)

3. Total initial deploy + cash reserve

Initial ticker basket (this strategy, founder click-through gates)

Total initial: 3.0R = $15,000. Cash reserve: $5,000 for Q2 adds.

Watchlist deltas — proposed for 2026 Q2 13F rebalance (added 2026-05-18, founder-approved)

Per Wright-synthesis concept article 2026-05-18 ([[../../../06-reference/concepts/2026-05-18-wright-agentic-capital-stack-vs-rdco-synthesis.md]]) — 3 named adds + 1 trigger-rule change. These do NOT deploy today; they take effect at the next 13F rebalance (mid-Aug 2026):

Explicit non-adds (documented 2026-05-18 to prevent re-litigation):

• Apollo / Ares — they are the BUYER class in Wright's frame, not the issuer/asset. Investing in Apollo doesn't capture the agentic capital markets thesis cleanly; we'd be paying for Apollo's existing private-credit franchise, not the agent-firm exposure.
• Sierra / Harvey — Wright's exemplar agent firms ($15B / $11B private valuations) but PRIVATE; can't be paper-traded on Alpaca. Move to private-market-watchlist if/when an SPV becomes available; otherwise watchlist-only for IPO timing.

Action timing: Ray drafts 2026 Q2 rebalance proposal post-2026-08-14 13F filings, surfaces via channel + /decisions/ page, founder approves the batch, Ray executes per the standard quarterly cadence below.

Rebalance rules (mechanical, quarterly)

Trigger window: 13F filing window (~45-60 days after quarter end). Calendar dates:

Rebalance actions (mechanical, executed by Ray after quarterly review with founder):

Quarterly review with founder before executing trims/exits. Ray drafts the rebalance proposal post-13F-filing window, surfaces via channel or /decisions/ page, founder approves the batch, Ray executes.

NO price-based exits

Same lesson as memory-cycle v1.1: drawdowns do not trigger sells. Only multi-manager exits trigger sells. A 30% drawdown on a persistent darling that 3 managers still hold = HOLD. A flat position that 2+ managers exited last quarter = CLOSE.

This is the strategy's core discipline: the mirror is the signal. If the people we're mirroring still hold it, so do we. If they bail, so do we (with a 45-90 day lag). Price action between disclosure windows is noise.

The single override: founder kill-switch via channel ("pause smart-money", "halt mirror") = close all immediately.

Audit + log shape

Every executed paper trade logs to:

~/rdco-vault/01-projects/investing/positions/smart-money-mirror-v1/
  <YYYY-MM-DD>-<ticker>-<side>.md

Per-trade frontmatter:

---
date: YYYY-MM-DD
strategy: smart-money-mirror-v1
ticker: <SYMBOL>
side: buy | sell | add | trim | close
size_r: 0.5
size_usd: 2500
price: <fill price>
signal_source: persistent-darling | multi-manager-add | multi-manager-exit | persistence-break
signal_managers: [tepper, druckenmiller, tiger-global]
signal_quarter: 2026Q1
trigger: initial-deploy | quarterly-rebalance | manual-override
notes: <free text>
---

Aggregate ledger maintained at positions/smart-money-mirror-v1/_ledger.md — running P/L, current R-deployment by ticker, cash reserve.

Quarterly review template

Filed to ~/rdco-vault/01-projects/investing/strategies/reviews/<YYYY-Q[N]>-smart-money-mirror-v1.md after each 13F window.

Checks each quarter:

1. New collective additions — any new 2+ tracked-manager same-quarter adds? List ticker, managers, and proposed add size.
2. New collective exits — any 2+ tracked-manager same-quarter exits on names we hold? List ticker, managers, proposed close.
3. Persistence delta on darlings — for each of MSFT / AMZN / GOOGL / TSM, count current top-10 managers. Flag any drop from 3+ → 2 (trim) or → 1 (close).
4. Cash reserve state — current cash, proposed deployment if any qualifying signal exists.
5. Total P/L vs SPY benchmark — strategy total return since inception vs SPY total return same window. The mirror should be at least beating SPY net of disclosure lag; if it isn't after 4+ quarters, the strategy needs revisit.
6. Anomalies / single-manager rotation noise — flag any names where 1 tracked manager moved hard but no second-manager confirmation. NOT a trade signal in v1; noted for future-quarter watch.

Disqualifying conditions (strategy-archive triggers)

The strategy retires (not just any single position closes) under these conditions:

Caveats / open risks

• Disclosure lag is 45-90 days. By the time we see a position, the manager may have already trimmed or exited. v1 accepts this; v2 could add a CUSIP-to-real-time-price layer to detect price moves between filing and our entry that materially change the read.
• Top-10 by dollar value is the wrong proxy in some cases. Berkshire's #1 by dollar (AAPL) has been trimmed substantially. Druckenmiller's top-10 churn is the highest. v1 uses top-10 inclusion as the binary signal; v2 could weight by position-as-%-of-portfolio.
• CUSIP-to-ticker mapping is not yet built. Per the backfill brief, all references use issuer-name strings. The v1 ticker basket is mapped manually (MSFT, AMZN, GOOGL, TSM, SNDK, INTC are unambiguous). Future strategy additions need the CUSIP cache before automation can stretch further.
• Pabrai excluded. The 8th tracked manager (Mohnish Pabrai) failed 13F backfill due to entity-name churn. v1 deploys without his signal; v2 could re-incorporate after the multi-entity CIK lookup is fixed.
• Marathon LLP filings stopped in 2021. The legendary UK capital-cycle firm appears below the $100M 13F threshold. Marathon-style capital-cycle reads via 13F are no longer possible — read Marathon's published 'Global Investment Review' commentary as substitute. Not blocking for v1.
• Berkshire 2025-Q2 returned only 4 positions (likely a 13F-HR/A amendment). The strict period-match logic misses the full Q2 holdings. Documented in the backfill brief; not blocking for v1 selection.
• Persistent-darling category is currently 100% mega-cap tech. No diversification across sector or market cap in the persistent slice. By design — that's where 3+ tracked managers converge — but it concentrates strategy beta into a single factor (AI-infra demand). Acknowledged.
• Overlap with memory-cycle v1.1. SNDK + INTC appear in both strategies' baskets. Position-tracking stays distinct per strategy (each gets its own ledger entry), but founder should be aware that total paper exposure to these two names is the sum across strategies.

Cross-domain translation

The shape here is mechanical mirror trading — also known in the literature as "13F arbitrage" or "smart-money following." Academic studies (Cohen/Polk/Silli 2010, Wermers 2000) showed equal-weighted top-N holdings of skilled active managers can beat passive benchmarks net of lag, but only when (a) the signal source is genuinely skilled, (b) holding period is long enough that the disclosure lag doesn't dominate, and (c) the basket is concentrated enough to capture the signal but diversified enough to survive single-manager error.

v1 satisfies (a) via the 7-manager curation (all positive-sum allocators with public track records). v1 satisfies (b) via the no-price-exit rule (holding period is "until a 2+ manager exit fires," not "until N days"). v1 partially satisfies (c) via the 6-name basket — concentrated enough to express the signal, diversified enough that any single manager's error costs <17% of the strategy.

The Sequoia Fund 2020-2022 analog: Sequoia ran a quarterly-rebalance concentrated portfolio with a similar shape (high-conviction names from a small manager curation, no mechanical price-based exits, quarterly review). The Druckenmiller doctrine ("find the great opportunity, bet heavily, let winners run") is the trading discipline that matches the mirror logic — we're not trying to time within positions, we're trying to ride positions that skilled allocators chose to size up on.

Comparison vs memory-cycle v1.1 (sibling strategy)

The two strategies are complementary, not competitive. Memory v1.1 is conviction-sized on a single structural thesis with the option to scale to 3R per name. Smart-money-mirror v1 is diversified across 6 mirror signals with smaller per-name caps and a tighter exit rule. Overlap (SNDK + INTC) is acknowledged; each strategy tracks its own positions.

Related

• [[01-projects/investing/README]] — operating model + R-unit framework
• [[01-projects/investing/anchors/smart-money/2026-05-17-aggregate-2yr-backfill-summary]] — selection-source brief
• [[2026-05-18-memory-cycle-v1.1]] — sibling thesis (memory bucket)
• [[.claude/skills/investing-smart-money-watch/SKILL.md]] — SOP for the quarterly 13F cron
• [[feedback_calibrate_overconfidence]] — applies: this strategy intentionally adds zero alpha vs the source signals; we're not out-thinking the managers, we're following them with discipline
• [[feedback_no_claudemd_state_drift]] — applies: strategy is locked here; rebalances file to reviews/ subfolder, not back into this doc

Open companion deliverables

• /decisions/2026-05-18-deploy-smart-money-mirror-v1-go.html — paper-deploy decision page (4-option click-back: APPROVE / REVISE / ARCHIVE / DEFER)
• _decisions.json entry — surfaces on HQ decisions index
• positions/smart-money-mirror-v1/ ledger folder — created on first executed trade post-approval
• Next quarterly review queued for ~2026-08-22 (post 2026 Q2 13F window)

Changelog

• 2026-05-18 (v1) — Initial strategy spec. Selection rules: persistent darlings (3+ managers, 2+ quarters) + recent multi-manager additions (2+ managers same quarter, last 2 quarters). Initial basket: MSFT / AMZN / GOOGL / TSM at 0.5R each + SNDK / INTC at 0.5R each = $15k initial deploy, $5k cash reserve. No price-based exits; quarterly rebalance on 13F filings; only mechanical exit is 2+ manager same-quarter exit. Pabrai excluded pending multi-entity CIK fix. Pending founder deploy gate via /decisions/ page.